TCP – Trajectory-Guided Control Prediction (Reproduced Checkpoint)
This model rehosts a community-trained reproduction of TCP: Trajectory-Guided Control Prediction for End-to-End Autonomous Driving (Wu et al., NeurIPS 2022).
It is provided only to enable reproducible evaluation in a bachelor’s-thesis context.
- Framework: PyTorch
- Input: front RGB camera (3×256×256) + target point (2)
- Output: continuous controls [steer, throttle, brake]
- Training data: CARLA 9.10 expert dataset
- File:
best_model.ckpt - License: Apache 2.0
Origin: unofficial reproduction shared in the public TCP GitHub issue #11 (Oct 2022).
Not affiliated with or endorsed by the original authors or OpenDriveLab.
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support